Skip to content

TFX-BSL Public Beam

tfx_bsl.public.beam

Module level imports for tfx_bsl.beam.

Functions

RunInference 

RunInference(
    examples: PCollection,
    inference_spec_type: InferenceSpecType,
    load_override_fn: Optional[LoadOverrideFnType] = None,
) -> PCollection

Run inference with a model.

There are two types of inference you can perform using this PTransform
  1. In-process inference from a SavedModel instance. Used when saved_model_spec field is set in inference_spec_type.
  2. Remote inference by using a service endpoint. Used when ai_platform_prediction_model_spec field is set in inference_spec_type.

examples: A PCollection containing examples of the following possible kinds, each with their corresponding return type. - PCollection[Example] -> PCollection[PredictionLog] * Works with Classify, Regress, MultiInference, Predict and RemotePredict.

  - PCollection[SequenceExample]           -> PCollection[PredictionLog]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[bytes]                     -> PCollection[PredictionLog]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

  - PCollection[Tuple[K, Example]]         -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * Works with Classify, Regress, MultiInference, Predict and
        RemotePredict.

  - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[Tuple[K, bytes]]           -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

inference_spec_type: Model inference endpoint. load_override_fn: Optional function taking a model path and sequence of tags, and returning a tf SavedModel. The loaded model must be equivalent in interface to the model that would otherwise be loaded. It is up to the caller to ensure compatibility. This argument is experimental and subject to change.

A PCollection (possibly keyed) containing prediction logs.

Source code in tfx_bsl/public/beam/run_inference.py 
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
@beam.ptransform_fn
@beam.typehints.with_input_types(_MaybeKeyedInput)
@beam.typehints.with_output_types(Union[_OUTPUT_TYPE, Tuple[_K, _OUTPUT_TYPE]])
def RunInference(
    examples: beam.pvalue.PCollection,
    inference_spec_type: model_spec_pb2.InferenceSpecType,
    load_override_fn: Optional[run_inference.LoadOverrideFnType] = None,
) -> beam.pvalue.PCollection:
    """Run inference with a model.

    There are two types of inference you can perform using this PTransform:
      1. In-process inference from a SavedModel instance. Used when
        `saved_model_spec` field is set in `inference_spec_type`.
      2. Remote inference by using a service endpoint. Used when
        `ai_platform_prediction_model_spec` field is set in
        `inference_spec_type`.

    Args:
    ----
      examples: A PCollection containing examples of the following possible kinds,
        each with their corresponding return type.
          - PCollection[Example]                   -> PCollection[PredictionLog]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[SequenceExample]           -> PCollection[PredictionLog]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[bytes]                     -> PCollection[PredictionLog]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

          - PCollection[Tuple[K, Example]]         -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[Tuple[K, bytes]]           -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

      inference_spec_type: Model inference endpoint.
      load_override_fn: Optional function taking a model path and sequence of
        tags, and returning a tf SavedModel. The loaded model must be equivalent
        in interface to the model that would otherwise be loaded. It is up to the
        caller to ensure compatibility. This argument is experimental and subject
        to change.

    Returns:
    -------
      A PCollection (possibly keyed) containing prediction logs.
    """
    return examples | "RunInferenceImpl" >> run_inference.RunInferenceImpl(
        inference_spec_type, load_override_fn
    )

Modules

run_inference

Public API of batch inference.

Functions
CreateModelHandler 
CreateModelHandler(
    inference_spec_type: InferenceSpecType,
) -> ModelHandler

Creates a Beam ModelHandler based on the inference spec type.

There are two model handlers
  1. In-process inference from a SavedModel instance. Used when saved_model_spec field is set in inference_spec_type.
  2. Remote inference by using a service endpoint. Used when ai_platform_prediction_model_spec field is set in inference_spec_type.

Example Usage:

from apache_beam.ml.inference import base

tf_handler = CreateModelHandler(inference_spec_type)
# unkeyed
base.RunInference(tf_handler)

# keyed
base.RunInference(base.KeyedModelHandler(tf_handler))

inference_spec_type: Model inference endpoint.

A Beam RunInference ModelHandler for TensorFlow

Source code in tfx_bsl/public/beam/run_inference.py 
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
def CreateModelHandler(
    inference_spec_type: model_spec_pb2.InferenceSpecType,
) -> ModelHandler:
    """Creates a Beam ModelHandler based on the inference spec type.

    There are two model handlers:
      1. In-process inference from a SavedModel instance. Used when
        `saved_model_spec` field is set in `inference_spec_type`.
      2. Remote inference by using a service endpoint. Used when
        `ai_platform_prediction_model_spec` field is set in
        `inference_spec_type`.

    Example Usage:

      ```
      from apache_beam.ml.inference import base

      tf_handler = CreateModelHandler(inference_spec_type)
      # unkeyed
      base.RunInference(tf_handler)

      # keyed
      base.RunInference(base.KeyedModelHandler(tf_handler))
      ```

    Args:
    ----
      inference_spec_type: Model inference endpoint.

    Returns:
    -------
      A Beam RunInference ModelHandler for TensorFlow
    """
    return run_inference.create_model_handler(inference_spec_type, None, None)
RunInference 
RunInference(
    examples: PCollection,
    inference_spec_type: InferenceSpecType,
    load_override_fn: Optional[LoadOverrideFnType] = None,
) -> PCollection

Run inference with a model.

There are two types of inference you can perform using this PTransform
  1. In-process inference from a SavedModel instance. Used when saved_model_spec field is set in inference_spec_type.
  2. Remote inference by using a service endpoint. Used when ai_platform_prediction_model_spec field is set in inference_spec_type.

examples: A PCollection containing examples of the following possible kinds, each with their corresponding return type. - PCollection[Example] -> PCollection[PredictionLog] * Works with Classify, Regress, MultiInference, Predict and RemotePredict.

  - PCollection[SequenceExample]           -> PCollection[PredictionLog]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[bytes]                     -> PCollection[PredictionLog]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

  - PCollection[Tuple[K, Example]]         -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * Works with Classify, Regress, MultiInference, Predict and
        RemotePredict.

  - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[Tuple[K, bytes]]           -> PCollection[
                                                  Tuple[K, PredictionLog]]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

inference_spec_type: Model inference endpoint. load_override_fn: Optional function taking a model path and sequence of tags, and returning a tf SavedModel. The loaded model must be equivalent in interface to the model that would otherwise be loaded. It is up to the caller to ensure compatibility. This argument is experimental and subject to change.

A PCollection (possibly keyed) containing prediction logs.

Source code in tfx_bsl/public/beam/run_inference.py 
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
@beam.ptransform_fn
@beam.typehints.with_input_types(_MaybeKeyedInput)
@beam.typehints.with_output_types(Union[_OUTPUT_TYPE, Tuple[_K, _OUTPUT_TYPE]])
def RunInference(
    examples: beam.pvalue.PCollection,
    inference_spec_type: model_spec_pb2.InferenceSpecType,
    load_override_fn: Optional[run_inference.LoadOverrideFnType] = None,
) -> beam.pvalue.PCollection:
    """Run inference with a model.

    There are two types of inference you can perform using this PTransform:
      1. In-process inference from a SavedModel instance. Used when
        `saved_model_spec` field is set in `inference_spec_type`.
      2. Remote inference by using a service endpoint. Used when
        `ai_platform_prediction_model_spec` field is set in
        `inference_spec_type`.

    Args:
    ----
      examples: A PCollection containing examples of the following possible kinds,
        each with their corresponding return type.
          - PCollection[Example]                   -> PCollection[PredictionLog]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[SequenceExample]           -> PCollection[PredictionLog]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[bytes]                     -> PCollection[PredictionLog]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

          - PCollection[Tuple[K, Example]]         -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[Tuple[K, bytes]]           -> PCollection[
                                                          Tuple[K, PredictionLog]]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

      inference_spec_type: Model inference endpoint.
      load_override_fn: Optional function taking a model path and sequence of
        tags, and returning a tf SavedModel. The loaded model must be equivalent
        in interface to the model that would otherwise be loaded. It is up to the
        caller to ensure compatibility. This argument is experimental and subject
        to change.

    Returns:
    -------
      A PCollection (possibly keyed) containing prediction logs.
    """
    return examples | "RunInferenceImpl" >> run_inference.RunInferenceImpl(
        inference_spec_type, load_override_fn
    )
RunInferenceOnKeyedBatches 
RunInferenceOnKeyedBatches(
    examples: PCollection,
    inference_spec_type: InferenceSpecType,
    load_override_fn: Optional[LoadOverrideFnType] = None,
) -> PCollection

Run inference over pre-batched keyed inputs.

This API is experimental and may change in the future.

Supports the same inference specs as RunInference. Inputs must consist of a keyed list of examples, and outputs consist of keyed list of prediction logs corresponding by index.

examples: A PCollection of keyed, batched inputs of type Example, SequenceExample, or bytes. Each type support inference specs corresponding to the unbatched cases described in RunInference. Supports - PCollection[Tuple[K, List[Example]]] - PCollection[Tuple[K, List[SequenceExample]]] - PCollection[Tuple[K, List[Bytes]]] inference_spec_type: Model inference endpoint. load_override_fn: Optional function taking a model path and sequence of tags, and returning a tf SavedModel. The loaded model must be equivalent in interface to the model that would otherwise be loaded. It is up to the caller to ensure compatibility. This argument is experimental and subject to change.

A PCollection of Tuple[K, List[PredictionLog]].

Source code in tfx_bsl/public/beam/run_inference.py 
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
@beam.ptransform_fn
@beam.typehints.with_input_types(_KeyedBatchesInput)
@beam.typehints.with_output_types(Tuple[_K, List[_OUTPUT_TYPE]])
def RunInferenceOnKeyedBatches(
    examples: beam.pvalue.PCollection,
    inference_spec_type: model_spec_pb2.InferenceSpecType,
    load_override_fn: Optional[run_inference.LoadOverrideFnType] = None,
) -> beam.pvalue.PCollection:
    """Run inference over pre-batched keyed inputs.

    This API is experimental and may change in the future.

    Supports the same inference specs as RunInference. Inputs must consist of a
    keyed list of examples, and outputs consist of keyed list of prediction logs
    corresponding by index.

    Args:
    ----
      examples: A PCollection of keyed, batched inputs of type Example,
        SequenceExample, or bytes. Each type support inference specs corresponding
        to the unbatched cases described in RunInference. Supports
          - PCollection[Tuple[K, List[Example]]]
          - PCollection[Tuple[K, List[SequenceExample]]]
          - PCollection[Tuple[K, List[Bytes]]]
      inference_spec_type: Model inference endpoint.
      load_override_fn: Optional function taking a model path and sequence of
        tags, and returning a tf SavedModel. The loaded model must be equivalent
        in interface to the model that would otherwise be loaded. It is up to the
        caller to ensure compatibility. This argument is experimental and subject
        to change.

    Returns:
    -------
      A PCollection of Tuple[K, List[PredictionLog]].
    """
    return (
        examples
        | "RunInferenceOnKeyedBatchesImpl"
        >> run_inference.RunInferenceImpl(inference_spec_type, load_override_fn)
    )
RunInferencePerModel 
RunInferencePerModel(
    examples: PCollection,
    inference_spec_types: Iterable[InferenceSpecType],
    load_override_fn: Optional[LoadOverrideFnType] = None,
) -> PCollection

Vectorized variant of RunInference (useful for ensembles).

examples: A PCollection containing examples of the following possible kinds, each with their corresponding return type. - PCollection[Example] -> PCollection[ Tuple[PredictionLog, ...]] * Works with Classify, Regress, MultiInference, Predict and RemotePredict.

  - PCollection[SequenceExample]          -> PCollection[
                                               Tuple[PredictionLog, ...]]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[bytes]                    -> PCollection[
                                               Tuple[PredictionLog, ...]]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

  - PCollection[Tuple[K, Example]]        -> PCollection[
                                               Tuple[K,
                                                     Tuple[PredictionLog,
                                                           ...]]]
      * Works with Classify, Regress, MultiInference, Predict and
        RemotePredict.

  - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                               Tuple[K,
                                                     Tuple[PredictionLog,
                                                           ...]]]
      * Works with Predict and (serialized) RemotePredict.

  - PCollection[Tuple[K, bytes]]           -> PCollection[
                                               Tuple[K,
                                                     Tuple[PredictionLog,
                                                           ...]]]
      * For serialized Example: Works with Classify, Regress,
        MultiInference, Predict and RemotePredict.
      * For everything else: Works with Predict and RemotePredict.

inference_spec_types: A flat iterable of Model inference endpoints. Inference will happen in a fused fashion (ie without data materialization), sequentially across Models within a Beam thread (but in parallel across threads and workers). load_override_fn: Optional function taking a model path and sequence of tags, and returning a tf SavedModel. The loaded model must be equivalent in interface to the model that would otherwise be loaded. It is up to the caller to ensure compatibility. This argument is experimental and subject to change.

A PCollection (possibly keyed) containing a Tuple of prediction logs. The Tuple of prediction logs is 1-1 aligned with inference_spec_types.

Source code in tfx_bsl/public/beam/run_inference.py 
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
@beam.ptransform_fn
@beam.typehints.with_input_types(_MaybeKeyedInput)
@beam.typehints.with_output_types(
    Union[Tuple[_OUTPUT_TYPE, ...], Tuple[_K, Tuple[_OUTPUT_TYPE, ...]]]
)
def RunInferencePerModel(
    examples: beam.pvalue.PCollection,
    inference_spec_types: Iterable[model_spec_pb2.InferenceSpecType],
    load_override_fn: Optional[run_inference.LoadOverrideFnType] = None,
) -> beam.pvalue.PCollection:
    """Vectorized variant of RunInference (useful for ensembles).

    Args:
    ----
      examples: A PCollection containing examples of the following possible kinds,
        each with their corresponding return type.
          - PCollection[Example]                  -> PCollection[
                                                       Tuple[PredictionLog, ...]]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[SequenceExample]          -> PCollection[
                                                       Tuple[PredictionLog, ...]]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[bytes]                    -> PCollection[
                                                       Tuple[PredictionLog, ...]]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

          - PCollection[Tuple[K, Example]]        -> PCollection[
                                                       Tuple[K,
                                                             Tuple[PredictionLog,
                                                                   ...]]]
              * Works with Classify, Regress, MultiInference, Predict and
                RemotePredict.

          - PCollection[Tuple[K, SequenceExample]] -> PCollection[
                                                       Tuple[K,
                                                             Tuple[PredictionLog,
                                                                   ...]]]
              * Works with Predict and (serialized) RemotePredict.

          - PCollection[Tuple[K, bytes]]           -> PCollection[
                                                       Tuple[K,
                                                             Tuple[PredictionLog,
                                                                   ...]]]
              * For serialized Example: Works with Classify, Regress,
                MultiInference, Predict and RemotePredict.
              * For everything else: Works with Predict and RemotePredict.

      inference_spec_types: A flat iterable of Model inference endpoints.
        Inference will happen in a fused fashion (ie without data
        materialization), sequentially across Models within a Beam thread (but
        in parallel across threads and workers).
      load_override_fn: Optional function taking a model path and sequence of
        tags, and returning a tf SavedModel. The loaded model must be equivalent
        in interface to the model that would otherwise be loaded. It is up to the
        caller to ensure compatibility. This argument is experimental and subject
        to change.

    Returns:
    -------
      A PCollection (possibly keyed) containing a Tuple of prediction logs. The
      Tuple of prediction logs is 1-1 aligned with inference_spec_types.
    """
    return (
        examples
        | "RunInferencePerModelImpl"
        >> run_inference.RunInferencePerModelImpl(
            inference_spec_types, load_override_fn
        )
    )
RunInferencePerModelOnKeyedBatches 
RunInferencePerModelOnKeyedBatches(
    examples: PCollection,
    inference_spec_types: Iterable[InferenceSpecType],
    load_override_fn: Optional[LoadOverrideFnType] = None,
) -> PCollection

Run inference over pre-batched keyed inputs on multiple models.

This API is experimental and may change in the future.

Supports the same inference specs as RunInferencePerModel. Inputs must consist of a keyed list of examples, and outputs consist of keyed list of prediction logs corresponding by index.

examples: A PCollection of keyed, batched inputs of type Example, SequenceExample, or bytes. Each type support inference specs corresponding to the unbatched cases described in RunInferencePerModel. Supports - PCollection[Tuple[K, List[Example]]] - PCollection[Tuple[K, List[SequenceExample]]] - PCollection[Tuple[K, List[Bytes]]] inference_spec_types: A flat iterable of Model inference endpoints. Inference will happen in a fused fashion (ie without data materialization), sequentially across Models within a Beam thread (but in parallel across threads and workers). load_override_fn: Optional function taking a model path and sequence of tags, and returning a tf SavedModel. The loaded model must be equivalent in interface to the model that would otherwise be loaded. It is up to the caller to ensure compatibility. This argument is experimental and subject to change.

A PCollection containing Tuples of a key and lists of batched prediction logs from each model provided in inference_spec_types. The Tuple of batched prediction logs is 1-1 aligned with inference_spec_types. The individual prediction logs in the batch are 1-1 aligned with the rows of data in the batch key.

Source code in tfx_bsl/public/beam/run_inference.py 
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
@beam.ptransform_fn
@beam.typehints.with_input_types(_KeyedBatchesInput)
@beam.typehints.with_output_types(Tuple[_K, Tuple[List[_OUTPUT_TYPE]]])
def RunInferencePerModelOnKeyedBatches(
    examples: beam.pvalue.PCollection,
    inference_spec_types: Iterable[model_spec_pb2.InferenceSpecType],
    load_override_fn: Optional[run_inference.LoadOverrideFnType] = None,
) -> beam.pvalue.PCollection:
    """Run inference over pre-batched keyed inputs on multiple models.

    This API is experimental and may change in the future.

    Supports the same inference specs as RunInferencePerModel. Inputs must consist
    of a keyed list of examples, and outputs consist of keyed list of prediction
    logs corresponding by index.

    Args:
    ----
      examples: A PCollection of keyed, batched inputs of type Example,
        SequenceExample, or bytes. Each type support inference specs corresponding
        to the unbatched cases described in RunInferencePerModel. Supports -
        PCollection[Tuple[K, List[Example]]] - PCollection[Tuple[K,
        List[SequenceExample]]] - PCollection[Tuple[K, List[Bytes]]]
      inference_spec_types: A flat iterable of Model inference endpoints.
        Inference will happen in a fused fashion (ie without data
        materialization), sequentially across Models within a Beam thread (but in
        parallel across threads and workers).
      load_override_fn: Optional function taking a model path and sequence of
        tags, and returning a tf SavedModel. The loaded model must be equivalent
        in interface to the model that would otherwise be loaded. It is up to the
        caller to ensure compatibility. This argument is experimental and subject
        to change.

    Returns:
    -------
      A PCollection containing Tuples of a key and lists of batched prediction
      logs from each model provided in inference_spec_types. The Tuple of batched
      prediction logs is 1-1 aligned with inference_spec_types. The individual
      prediction logs in the batch are 1-1 aligned with the rows of data in the
      batch key.
    """
    return (
        examples
        | "RunInferencePerModelOnKeyedBatchesImpl"
        >> run_inference.RunInferencePerModelImpl(
            inference_spec_types, load_override_fn
        )
    )
Modules